Electrical control of quantum bits in silicon paves the way to large quantum computers

A Univ. of New South Wales (UNSW)-led research team has encoded quantum information in silicon using simple electrical pulses for the first time, bringing the construction of affordable large-scale quantum computers one step closer to reality.

Lead researcher, UNSW Assoc. Prof. Andrea Morello from the School of Electrical Engineering and Telecommunications, said his team had successfully realized a new control method for future quantum computers.

The findings were published in Science Advances.

Unlike conventional computers that store data on transistors and hard drives, quantum computers encode data in the quantum states of microscopic objects called qubits.

The UNSW team, which is affiliated with the ARC Centre of Excellence for Quantum Computation & Communication Technology, was first in the world to demonstrate single-atom spin qubits in silicon, reported in Nature in 2012 and 2013.

Team tightens bounds on quantum information “speed limit”

If you're designing a new computer, you want it to solve problems as fast as possible. Just how fast is possible is an open question when it comes to quantum computers, but physicists at NIST have narrowed the theoretical limits for where that "speed limit" is. The research implies that quantum processors will work more slowly than some research has suggested.

The work offers a better description of how quickly information can travel within a system built of quantum particles such as a group of individual atoms. Engineers will need to know this to build quantum computers, which will have vastly different designs and be able to solve certain problems much more easily than the computers of today. While the new finding does not give an exact speed for how fast information will be able to travel in these as-yet-unbuilt computers—a longstanding question—it does place a far tighter constraint on where this speed limit could be.



Quantum computers will store data in a particle's quantum states—one of which is its spin, the property that confers magnetism. A quantum processor could suspend many particles in space in close proximity, and computing would involve moving data from particle to particle. Just as one magnet affects another, the spin of one particle influences its neighbor's, making quantum data transfer possible, but a big question is just how fast this influence can work.

The NIST team's findings advance a line of research that stretches back to the 1970s, when scientists discovered a limit on how quickly information could travel if a suspended particle only could communicate directly with its next-door neighbors. Since then, technology advanced to the point where scientists could investigate whether a particle might directly influence others that are more distant, a potential advantage. By 2005, theoretical studies incorporating this idea had increased the speed limit dramatically.

"Those results implied a quantum computer might be able to operate really fast, much faster than anyone had thought possible," says NIST's Michael Foss-Feig. "But over the next decade, no one saw any evidence that the information could actually travel that quickly."

Physicists exploring this aspect of the quantum world often line up several particles and watch how fast changing the spin of the first particle affects the one farthest down the line—a bit like standing up a row of dominoes and knocking the first one down to see how fast the chain reaction takes. The team looked at years of others' research and, because the dominoes never seemed to fall as fast as the 2005 prediction suggested, they developed a new mathematical proof that reveals a much tighter limit on how fast quantum information can propagate.

"The tighter a constraint we have, the better, because it means we'll have more realistic expectations of what quantum computers can do," says Foss-Feig.

The limit, their proof indicates, is far closer to the speed limits suggested by the 1970s result.

The proof addresses the rate at which entanglement propagates across quantum systems. Entanglement—the weird linkage of quantum information between two distant particles—is important, because the more quickly particles grow entangled with one another, the faster they can share data. The 2005 results indicated that even if the interaction strength decays quickly with distance, as a system grows, the time needed for entanglement to propagate through it grows only logarithmically with its size, implying that a system could get entangled very quickly. The team's work, however, shows that propagation time grows as a power of its size, meaning that while quantum computers may be able to solve problems that ordinary computers find devilishly complex, their processors will not be speed demons.

Graphics in reverse

Most recent advances in artificial intelligence—such as mobile apps that convert speech to text—are the result of machine learning, in which computers are turned loose on huge data sets to look for patterns.

To make machine-learning applications easier to build, computer scientists have begun developing so-called probabilistic programming languages, which let researchers mix and match machine-learning techniques that have worked well in other contexts. In 2013, the U.S. Defense Advanced Research Projects Agency (DARPA), an incubator of cutting-edge technology, launched a four-year program to fund probabilistic-programming research.



At the Computer Vision and Pattern Recognition conference in June, Massachusetts Institute of Technology (MIT) researchers will demonstrate that on some standard computer-vision tasks, short programs—less than 50 lines long—written in a probabilistic programming language are competitive with conventional systems with thousands of lines of code.

“This is the first time that we’re introducing probabilistic programming in the vision area,” says Tejas Kulkarni, an MIT graduate student in brain and cognitive sciences and first author on the new paper. “The whole hope is to write very flexible models, both generative and discriminative models, as short probabilistic code, and then not do anything else. General-purpose inference schemes solve the problems.”

By the standards of conventional computer programs, those “models” can seem absurdly vague. One of the tasks that the researchers investigate, for instance, is constructing a 3-D model of a human face from 2-D images. Their program describes the principal features of the face as being two symmetrically distributed objects (eyes) with two more centrally positioned objects beneath them (the nose and mouth). It requires a little work to translate that description into the syntax of the probabilistic programming language, but at that point, the model is complete. Feed the program enough examples of 2-D images and their corresponding 3-D models, and it will figure out the rest for itself.

“When you think about probabilistic programs, you think very intuitively when you’re modeling,” Kulkarni says. “You don’t think mathematically. It’s a very different style of modeling.”

Joining Kulkarni on the paper are his adviser, professor of brain and cognitive sciences Josh Tenenbaum; Vikash Mansinghka, a research scientist in MIT’s Dept. of Brain and Cognitive Sciences; and Pushmeet Kohli of Microsoft Research Cambridge. For their experiments, they created a probabilistic programming language they call Picture, which is an extension of Julia, another language developed at MIT.

What’s old is new
The new work, Kulkarni says, revives an idea known as inverse graphics, which dates from the infancy of artificial-intelligence research. Even though their computers were painfully slow by today’s standards, the artificial intelligence pioneers saw that graphics programs would soon be able to synthesize realistic images by calculating the way in which light reflected off of virtual objects. This is, essentially, how Pixar makes movies.

Some researchers, like the MIT graduate student Larry Roberts, argued that deducing objects’ three-dimensional shapes from visual information was simply the same problem in reverse. But a given color patch in a visual image can, in principle, be produced by light of any color, coming from any direction, reflecting off of a surface of the right color with the right orientation. Calculating the color value of the pixels in a single frame of “Toy Story” is a huge computation, but it’s deterministic: All the variables are known. Inferring shape, on the other hand, is probabilistic: It means canvassing lots of rival possibilities and selecting the one that seems most likely.

That kind of inference is exactly what probabilistic programming languages are designed to do. Kulkarni and his colleagues considered four different problems in computer vision, each of which involves inferring the three-dimensional shape of an object from 2-D information. On some tasks, their simple programs actually outperformed prior systems. The error rate of the program that estimated human poses, for example, was between 50 and 80% lower than that of its predecessors.

Learning to learn
In a probabilistic programming language, the heavy lifting is done by the inference algorithm—the algorithm that continuously readjusts probabilities on the basis of new pieces of training data. In that respect, Kulkarni and his colleagues had the advantage of decades of machine-learning research. Built into Picture are several different inference algorithms that have fared well on computer-vision tasks. Time permitting, it can try all of them out on any given problem, to see which works best.

Moreover, Kulkarni says, Picture is designed so that its inference algorithms can themselves benefit from machine learning, modifying themselves as they go to emphasize strategies that seem to lead to good results. “Using learning to improve inference will be task-specific, but probabilistic programming may alleviate re-writing code across different problems,” he says. “The code can be generic if the learning machinery is powerful enough to learn different strategies for different tasks.”

Advances in molecular electronics

Scientists at the Helmholtz-Zentrum Dresden-Rossendorf (HZDR) and the Univ. of Konstanz are working on storing and processing information on the level of single molecules to create the smallest possible components that will combine autonomously to form a circuit. As recently reported in Advanced Science, the researchers can switch on the current flow through a single molecule for the first time with the help of light.

Dr. Artur Erbe, physicist at the HZDR, is convinced that in the future molecular electronics will open the door for novel and increasingly smaller—while also more energy efficient—components or sensors: "Single molecules are currently the smallest imaginable components capable of being integrated into a processor." Scientists have yet to succeed in tailoring a molecule so that it can conduct an electrical current and that this current can be selectively turned on and off like an electrical switch.

This requires a molecule in which an otherwise strong bond between individual atoms dissolves in one location—and forms again precisely when energy is pumped into the structure. Dr. Jannic Wolf, chemist at the Univ. of Konstanz, discovered through complex experiments that a particular diarylethene compound is an eligible candidate. The advantages of this molecule, approximately three nanometers in size, are that it rotates very little when a point in its structure opens and it possesses two nanowires that can be used as contacts. The diarylethene is an insulator when open and becomes a conductor when closed. It thus exhibits a different physical behavior, a behavior that the scientists from Konstanz and Dresden were able to demonstrate with certainty in numerous reproducible measurements for the first time in a single molecule.



A computer from a test-tube
A special feature of these molecular electronics is that they take place in a fluid within a test-tube, where the molecules are contacted within the solution. In order to ascertain what effects the solution conditions have on the switching process, it was therefore necessary to systematically test various solvents. The diarylethene needs to be attached at the end of the nanowires to electrodes so that the current can flow. "We developed a nanotechnology at the HZDR that relies on extremely thin tips made of very few gold atoms. We stretch the switchable diarylethene compound between them," explains Dr. Erbe.

When a beam of light then hits the molecule, it switches from its open to its closed state, resulting in a flowing current. "For the first time ever we could switch on a single contacted molecule and prove that this precise molecule becomes a conductor on which we have used the light beam," says Dr. Erbe, pleased with the results. "We have also characterized the molecular switching mechanism in extremely high detail, which is why I believe that we have succeeded in making an important step toward a genuine molecular electronic component."

Switching off, however, does not yet work with the contacted diarylethene, but the physicist is confident: "Our colleagues from the HZDR theory group are computing how precisely the molecule must rotate so that the current is interrupted. Together with the chemists from Konstanz, we will be able to accordingly implement the design and synthesis for the molecule." However, a great deal of patience is required because it's a matter of basic research. The diarylethene molecule contact using electron-beam lithography and the subsequent measurements alone lasted three long years. Approximately ten years ago, a working group at the Univ. of Groningen in the Netherlands had already managed to construct a switch that could interrupt the current. The off-switch also worked only in one direction, but what couldn't be proven at the time with certainty was that the change in conductivity was bound to a single molecule.

Nano-electronics in Dresden
One area of research focus in Dresden is what is known as self-organization. "DNA molecules are, for instance, able to arrange themselves into structures without any outside assistance. If we succeed in constructing logical switches from self-organizing molecules, then computers of the future will come from test-tubes," Dr. Erbe prophesizes. The enormous advantages of this new technology are obvious: billion-euro manufacturing plants that are necessary for manufacturing today's microelectronics could be a thing of the past. The advantages lie not only in production but also in operating the new molecular components, as they both will require very little energy.

With the Helmholtz Research School NANONET, the conditions for investigating and developing the molecular electronics of tomorrow are quite positive in Dresden. In addition to the HZDR, the Technische Universität Dresden, Leibniz-Institute of Polymer Research Dresden (IPF), the Fraunhofer Institute for Ceramic Technology and Systems (IKTS) and the NaMLab gGmbH all participate in running the structured doctoral program.

Breaking Down Barriers: Streamlining Data Management to Boost Knowledge Sharing

Research in the pharmaceutical and industrial science industries has become increasingly global, multidisciplinary and data-intensive. This is made clear by the evolution in patent approvals, which can also be considered a reliable measure of innovation in these industries. Innovation itself, of course, is a cumulative effect, which requires access to multiple fragments of knowledge from disparate sources and exchange of technology and ideas.

While the benefits in innovation in such a competitive environment are clear, investment in research is primarily influenced by the strategic behavior of companies, and a deeper understanding of the importance of market share. Patents and publications help to establish corporate reputation, allowing for controlled technology transfer with strategic joint ventures and to raise barriers to prevent competitors from eroding market share.



The relationship between technological processes, innovation and economic growth has changed over time, as innovation and technological advancement became increasingly important for sustained economic performance. This change was largely driven by globalization, with concurrent flows of information, technology, capital and services and resources across the world, and was manifested by the rising investment in market-oriented research, a surge in patenting driven by rapid innovation across all technology fields and a broad investment in the services sectors.

For those who fund the research, the sharing—and therefore efficient use—of data is a high priority. This keeps the “knowledge management” cogs turning, helping organizations to create, acquire, disseminate and leverage knowledge in order to retain competitive advantage. In R&D, this process increasingly requires researchers to externalize and exchange information, to increase the productivity and profitability of the organization. This growing emphasis on knowledge sharing is a significant, step-change in the way research is carried out—and presents new challenges to the R&D ecosystem.

Moving beyond the “paper prison”
Although efficient knowledge management and sharing is seen as key to increasing productivity and profitability of organizations, there are a number of potential barriers that can exist within an organization—primarily created by factors such as hierarchy, motivation, flexibility and transparency of the communication system within the organization.

Many researchers are familiar with the challenges of data storage, given that important research may often be archived in paper notebooks, computers, external hard drives and corporate IT systems. Although document management systems encompass enterprise storage capability for IP compliance, often these fail to capture the tacit knowledge of the researcher—and crucially, the context of how and why the data was created. The introduction of electronic laboratory notebooks has helped to overcome this, by providing an environment that allows the researcher to capture the experimental design process, together with the data and conclusions as the experiment is conducted.

Addressing the human factors
Trust is an important influence on an individual’s reticence to share knowledge. Employees may believe they are in competition with each other, and that the action of sharing knowledge may result in them losing power and influence in the organization. Employees may also not be willing to share information unless they are sure their knowledge is safe from misuse, or that they are certain about the results. Traditionally, such information may have been controlled by visibility and access to the paper notebook where the information was stored. In an electronic laboratory notebook, private areas can be created to hide data from public view, until an experiment has been completed and the results have been validated. Equally, these protected areas may be created to protect sensitive data, or to segment in-house research from that conducted by a contract research organization.

Human capital is an important component of the innovation process, and requires a deeper understanding the soft skills of teamwork and inter-personal relationships. Communication skills and knowledge transfer of employees are thus positively influenced by the level of interaction within the organization (given the opportunity, distance and visibility of the channel of interaction within an organization), but may be equally challenged by a “know-it-all” attitude, poor ability to comprehend the information being exchanged or a fear of receiving negative criticism.

Tackling the infrastructure obstacles
From an organizational perspective, barriers may also exist due to ethnographic language differences—particularly prevalent in global organizations—or where inherent differences in culture exist because of successive mergers and acquisitions. It is said that one of the key challenges to successful mergers of organizations is to reconcile and adopt a new organizational culture, but this may take years for managers to effectively develop and implement successfully.

Organizations can help to overcome this by creating a recognition system to reward employees for sharing information, or by accrediting those whose work contributes to new patents and publications. By enabling information exchange sessions between remote teams, an open culture of knowledge sharing may be established. The organization itself needs to recognize it’s cheaper to re-use information for both successful and failed research, than it is to repeat the work of someone else. In a paper notebook world, it’s almost impossible to identify what has been done by a co-worker in a foreign site. However with text mining of documents, an electronic data repository offers users a facile way to use keywords to search for data that’s analogous to their own research aims.

Technology also forms a key part of the knowledge management infrastructure, along with the employee resource and the processes of data capture. It forms the backbone of intra-organizational knowledge sharing, particularly where multiple research sites exist in different geographic locations. By connecting sites, the research operation become decentralized, although potential technology barriers may result from a lack of integration of the information systems, together with a disconnect between employees’ expectations of the technology and what it’s capable of delivering. Additionally, researchers now often work with a multitude of systems and instruments, and it’s important to recognize not all users have the same degree of capability or access to these. Although some of these barriers can be overcome through education or formal training, users may simply suffer from slow network speeds between sites, which can hinder system adoption and a willingness to search for prior research remotely.

Clearly, the R&D process is evolving. Firms must now manage and share knowledge, and deal with an evolving set of associated challenges in doing so—but these can be overcome. An open corporate culture, coupled with effective data management tools, helps to break the communication barrier by linking researchers across different geographies and business units. This ensures researchers are able collaborate effectively and reuse existing data, to seed new discoveries and keep science moving forwards.

Deadline Extended for 2015 R&D 100 Award Entries

The editors of R&D Magazine have announced a deadline extension for the 2015 R&D 100 Awards entry process until May 18, 2015.

The R&D 100 Awards have a 50 plus year history of awarding the 100 most technologically significant products of the year. Past winners have included sophisticated testing equipment, innovative new materials, chemistry breakthroughs, biomedical products, consumer items, high-energy physics and more. The R&D 100 Awards span industry, academia and government-sponsored research.

This year we have made the entry form shorter and simpler from the last year’s already overhauled version. That means less questions and more time to enter your products.

Register Now

What products qualify?
Any new technical product or process that was first available for purchase or licensing between January 1, 2014, and March 31, 2015, is eligible for the 2015 awards. This includes manufacturing processes such as machining, open source software, new types of materials or chemicals and consumer-level products such as cameras. Proof-of-concepts and early-stage prototypes don’t quality, however; the submitted entry must be in working, marketable condition.



For more information on the 2015 R&D 100 Award entry process, please visit www.rd100awards.com.